AGRICULTURAL
ECONOMICS
ELSEVIER
Agricultural Economics 22 (2000) 75-90
www.elsevier.com/locate/agecon
The economics of coupled farm subsidies under costly
and imperfect enforcement
Konstantinos Giannakas a,*, Murray Fulton b
• Department of Agricultural Economics, University of Nebraska-Lincoln, Lincoln, Nebraska, USA 68583-0922
b Department of Agricultural Economics and Centre for the Study of Co-operatives, University of Saskatchewan,
Saskatoon, Sask., Canada S7N 5A8
Received 30 January 1998; received in revised form 22 July 1999; accepted 31 August 1999
Abstract
This study relaxes the assumption of perfect and costless policy enforcement found in traditional agricultural policy
analysis and introduces enforcement costs and cheating into the economic analysis of output subsidies. Policy design and
implementation is modeled in this paper as a sequential game between the regulator who decides on the level of intervention,
an enforcement agency that determines the level of policy enforcement, and the farmer who makes the production and cheating
decisions. Analytical results show that farmer compliance is not the natural outcome of self-interest and complete deterrence of
cheating is not economically efficient. The analysis also shows that enforcement costs and cheating change the welfare effects
of output subsidies, the efficiency of the policy instrument in redistributing income, the level of government intervention that
transfers a given surplus to agricultural producers, the socially optimal income redistribution, and the social welfare from
intervention. ©2000 Elsevier Science B.V. All rights reserved.
JEL classification: H22; H23; K42; Q18
Keywords: Agricultural policy; Enforcement; Misrepresentation; Subsidies; Transfer efficiency
1. Introduction
The traditional analysis of production subsidies
(per unit and ad valorem subsidies) and deficiency
payments takes place under the assumption that either farmers comply completely with the provisions
of these farm programs or policy enforcement is perfect and costless. If cheating on farm programs is
* Corresponding author. Present address: Department of Agricultural Economics, University of Nebraska-Lincoln, 216 H.C. Filley
Hall, Lincoln, Nebraska, USA 68583-0922. Tel.: 1-402-472-2041;
fax: + 1-402-472-3460
E-mail address: [email protected] (K. Giannakas)
profitable however, full compliance is by no means
assured unless monitoring and enforcement are costlessly carried out. 1 Because of the cost associated
with investigating farmers and punishing the detected
cheaters, program enforcement is likely to be incom-
1 Monitoring refers to the actions taken by policy makers to determine whether a farmer complies with the provisions of the subsidy program and includes on-site inspections and reviewing the
records of a farm. Enforcement refers to the process of moving
violators into compliance through penalties and/or criminal prosecution for instance. In what follows, the terms monitoring and/or
enforcement will be used interchangeably to denote the process
of investigating farmers and punishing the detected cheaters.
0169-5150/00/$- see front matter ©2000 Elsevier Science B.V. All rights reserved.
PII: SO 169-5150(99)00042-0
76
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
plete. Imperfect enforcement in turn creates economic
incentives for farmers to cheat on the farm program.
In fact, cases of cheating on farm programs are often
reported by the European Press (Moyer and Josling,
1990). Fraud is considered to be an integral part of
the common agricultural policy (CAP) (Ockenden and
Franklin, 1995; Gardner, 1996). Gardner argues that
the manipulation of regulations "to milk the maximum
in subsidies" from the European Union's (EU) farm
budget, "is not only common, standard practice... but
is also accepted as standard practice" (Gardner, 1996,
pp. 44).
The possibility of cheating on subsidy programs
in the EU arises from the fact that eligibility for
most government payments usually requires farmers to make an application for the payment (Harvey,
1997). 2 By over-reporting the level of their production on the application, farmers can collect payments
on quantities greater than those produced. A recent
report on the extent of cheating on farm subsidies in
the EU estimates the 'losses' through fraud and lax
controls in the payment of farm subsidies and subsidy
overpayments to $4 billion per year. 3 These cheating
costs account for up to 10% of the 36 billion ECU a
year laid out on agricultural support through the CAP
(Gardner, 1996).
Cheating on farm programs is not an exclusive European characteristic however. In the United States, the
existence of a United States Department of Agriculture (USDA) 'hotline', where cases of 'fraud' related
to 'submission of false claims/statements' can be reported, indicates that the problem of cheating on farm
programs is not unknown to US agricultural policy
makers (USDA Office of Inspector General, 1999).
Very few studies have incorporated misrepresentation or cheating in theoretical agricultural policy analysis. Among the exceptions are Alston and Smith
(1983) who raise the question of cheating and 'black
market' activity in an examination of rationing in an
industry with an effective minimum price policy in
place and Giannakas (1998) who examines the incidence of output quotas, deficiency payments and de2 There are complaints heard throughout Europe that farmers
spend more time filling in forms than farming (Ockenden and
Franklin, 1995).
3 EU fraud and waste on farm subsidies, 117 Journal of Commerce, 22 December 1997.
coupled area payments under costly and imperfect enforcement. 4
The purpose of this study is to introduce enforcement costs into the economic analysis of output subsidies (production subsidies and deficiency payments)
and to examine the causes and consequences of farmer
misrepresentation. The consequences of cheating for
the incidence of output subsidies are considered in
the context of a static, partial equilibrium model of a
closed economy. A key element of the subsidy policy modeled in this paper is that farmers report their
own production. Therefore, the results of the analysis
apply to cases where eligibility for government subsidies requires farmers to make an application for the
payment.
Analytical results show that compliance with policy
rules is not the expected producer behavior and complete deterrence of cheating is not economically efficient. The analysis also shows that enforcement costs
and cheating change the welfare effects of output subsidies, the efficiency of the policy instrument in transferring income to agricultural producers, the level of
government intervention that transfers a given surplus
to producers, the socially optimal income redistribution, and the social welfare from intervention. The relevance of the analysis for a small open economy and
the large country case is discussed throughout the text.
2. Agricultural policy making
The agricultural policy making structure in most
countries, and certainly the EU and the US, is characterized by a separation of powers between the agencies responsible for policy design and policy enforcement. In the EU, the level of the transfer is decided by
the Council of Ministers of Agriculture and the countries/members are assigned the implementation and
4 This paper is based on Chapter IV of Giannakas' Ph.D. Dissertation (Giannakas, 1998). Working papers currently in review
examine the other policy instruments considered in the dissertation, namely output quotas and decoupled area payments. A fourth
paper (Giannakas and Fulton, 1999) focuses on the efficiency of
quotas and subsidies in an institutional structure in which a single
agency determines both enforcement and policy intervention. The
assumption of a single agency is in contrast to the institutional
arrangement with separate regulatory and enforcement agencies
examined in this paper.
K. Giannakas, M. Fulton/Agricultural Economics 22 (2000) 75-90
enforcement of the policy (Runge and von Witzke,
1987; From and Stava, 1993; Swinbank, 1997). In the
US, the legislative body of the government (i.e., the
Congress) decides on the farm policy and the level
of transfer to producers and the USDA is responsible
for carrying out the programs (Gardner, 1987; Moyer
and Josling, 1990). More specifically, the Secretary of
Agriculture decides on the level of the policy instrument that achieves the policy objectives, and the Agricultural Stabilization and Conservation Service (now
the Farm Services Agency) implements and enforces
the commodity programs.
To capture this separation of activities, an institutional arrangement characterized by decentralized policy making is adopted in this paper. More specifically,
policy design and implementation is modeled in this
paper as a sequential game between the regulator who
decides on the level of intervention, an enforcement
agency that determines the level of policy enforcement, and the farmer who decides on the quantity to
produce and the quantity to misrepresent. The payoff functions of all agents are assumed to be common knowledge. The regulator moves first and decides on the subsidy knowing exactly how her decisions will affect enforcement and misrepresentation.
Optimal enforcement is determined next. Finally, the
farmer makes his production and cheating decisions,
observing both the policy variable and the enforcement parameters. Different scenarios concerning the
political preferences of the enforcement agency and
the decision variables it controls are examined within
this framework.
All formulations of the sequential game developed
in this paper are solved using backwards induction
(Kreps, 1990; Gibbons, 1992). The problem of the
farmer is considered first, the problem of the enforcement agency follows and, finally, the solution to the
problem of the regulatory agency determines optimal
intervention, enforcement and misrepresentation.
3. Output subsidies and optimal farmer
misrepresentation
Production subsidies and deficiency payments have
often been used by policy makers, alone or in conjunction with other policies like price supports, supply controls, and/or trade policies, to encourage production of
77
a specific commodity and/or to transfer income to producers. Both producers and consumers have benefited
from production subsidies and deficiency payments
while taxpayers have incurred the costs. Fig. 1 depicts the traditional static, partial equilibrium welfare
effects of a production subsidy scheme for a closed
economy with linear approximations of supply and demand curves (Nerlove, 1958; Wallace, 1962). In this
static context the production subsidy program is identically equivalent to a target price-deficiency payment
scheme. 5 In what follows, the terms output subsidy
and subsidy will be used to denote a production subsidy and/or a deficiency payment.
Under a per unit output subsidy of v, producers receive an increase in producer surplus equal to the area
ptBCpe, consumers gain area peCDpc, while taxpayers lose area (1 +d) lBDpc, where dis the marginal
deadweight loss from taxation (Ballard and Fullerton,
1992). Taxpayers' cost is given by the product of the
market cleruing quantity and the subsidy paid to farmers, adjusted to account for the positive deadweight
losses from taxation. The distortionary costs of market intervention equal the area BCD plus d(iBDpc)
(Gardner, 1983,1987; Alston and Hurd, 1990). The
implicit assumption in this analysis is that farmers do
not misrepresent their production, i.e., farmers do not
cheat.
Given the increased benefits from an output subsidy scheme, however, fanners may find it economically optimal to increase further their returns by
over-reporting the level of their production and collecting government payments for phantom output.
Assuming farmers know with certainty the subsidy for
their production, the penalty in case they are caught
cheating, and the probability of being investigated,
5 The equivalence of deficiency payments and production subsidies (i.e., per unit subsidies and ad valorem subsidies) may break
down when market conditions change. Since, by definition, deficiency payments make up the difference between some target
price and the corresponding market clearing price, changes in supply and/or demand leave the price received by producers in the
presence of the program unaffected; the deficiency payment is adjusted so that the (target) price is constant. On the other hand,
when a per unit subsidy (or an ad valorem subsidy) is in effect,
the price received by producers (and, thus, the quantity produced)
changes as the market conditions change; production subsidies do
not stabilize the price received by producers. Changes in the market conditions also change the per unit subsidy that is equivalent
to a particular ad valorem subsidy (Gardner, 1987).
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
78
p
s
~
--------- -------------------- B
pe ·····························
~
------------
-------
D
0~~------------------------~----------·
qt
Q
Fig. 1. The welfare effects of production subsidies and deficiency payments.
their decision on whether to cheat (and if so, by how
much) can be modeled as decision-making under uncertainty. In this framework, the individual farmer's
choice can be viewed as a choice between a certain
outcome (profits if he does not cheat) and his profits
in case he misrepresents his level of production. Assuming the representative farmer is risk neutral, his
objective function can be written as:
maXqt,qmE[IT]
= (pc + v)q 1 -
c(q 1)
+[(1-8)v-8p] qm
(1)
where pc is the market clearing price; v the output
subsidy; q1 quantity produced; c(q1) cost function; qm
is the quantity reported as eligible for payments over
and above q1 ; pis the per unit penalty charged on detected misrepresentation; and 8 is the probability that
the farmer will be audited. 6 If the farmer is cheat6 The model in Eq. (1) can be modified to include risk aversion
of the representative farmer and/or private costs from cheating.
The risk averse farmer will choose qm that maximizes his expected utility (i.e., maxq'.qmE[U(n)] = (1- 8)U[(pc + v)q 1 c(q 1) +vqm] +8U[(pc +v)q 1 -c(q 1) - pqm]). In terms of output
misrepresentation, risk aversion results in reduced cheating relative to the case where risk neutrality is assumed. Cheating also
ing on the farm program, 8 reflects the probability he
will be detected and punished. The probability of audit takes values between 0 and 1 (i.e., 8 E [0,1]) and
reflects the intensity with which agricultural policy is
enforced and with which cheating is investigated.
The audit probability is assumed to be a linear
function of the amount of cheating, i.e., 8 =8o + 81qm,
where 8o is a fixed base probability and 81 qm is a component that depends on the misrepresented quantity.
The parameter 8o is assumed to be dependent on the
resources spent for policy enforcement. The parameter 8 1 depends on factors affecting the observability of
farmers' actions (e.g., such as location and dispersion
of the farms) and is assumed to be strictly positive and
exogenous to both policy enforcers and producers.
falls when the costs incurred by farmers in protecting themselves
from detection (i.e., k(qm)) are incorporated into the representative
farmer's objective function. Even though both risk averse behavior and private costs from cheating change the results quantitatively, the qualitative nature of the results in this study remains
unaffected. Notice that in cases where government payments are
based on delivered product, cheating requires collusion between
the agency responsible for the payments and the farmer. In such a
case, the component kqm would represent the bribes to the corrupt
government official and/or the processor.
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
The first order conditions for the representative
farmer's problem are
aE [TI]
aqt
= 0 ::::} PC
+v =
c' (qt)
(2)
a
E [TIJ
- - - =0::::} v- 8o(v+ p)- 28J(V + p)qm
v-8o(v+p)
=0::::}qm=
28J(V + p)
to the origin of Din Fig. 3. The slopes of DELTA and
MPO curves are 81 IN and 281/N, respectively, where N
is the number of representative farmers producing the
subsidized commodity. The intersection of vi( v + p)
and MPO gives the aggregate quantity misrepresented
at equilibrium which can be written as
v- 8o(v + p)
Qm = N qm = -28_,i-=(v__:__+_p....:.)....:..
aqm
79
where
8i = ~
N
(4)
(3)
Eq. (2) shows that the farmer will produce where
the market price plus subsidy equals the marginal cost
of production. Note that the optimal output level does
not depend on any of the parameters associated with
farmer misrepresentation.
Eq. (3) shows the optimal choice of the quantity
misrepresented by the representative farmer as a function of subsidy payments, per unit penalty and audit
probability parameters. Eq. (3), thus, reflects the best
response of the farmer to the choices made by the regulatory and enforcing agencies. Consistent with a priori
expectations, misrepresented quantity increases with
the subsidy payment and decreases with an increase in
the audit probability and per unit penalty parameters
(i.e.,aqmfav > 0, aqmfa8o < 0 and aqmfap < 0).
Manipulation of the expression for qm indicates that
the over-reported quantity is positive when 8o is less
than vl(v+ p), when vis greater than 8op/(1- 8o), or
when pis less than(l-8o)v /(8o). These critical values
for 8o, v and are denoted 88c, vnc and pnc, respectively,
where the superscript nc stands for no cheating.
A manipulation of Eq. (3) shows that the optimal level of misrepresentation is given by equating
vl(v+ p) and 8o + 28Jqm. Fig. 2 shows this relationship graphically. The horizontal line vl(v+ p) shows
the ratio of the marginal benefits in case cheating
goes undetected, over the opportunity cost in case
the farmer is caught cheating. The line 8o + 28Jqm
shows the change in the output that is expected to be
penalized for a change in the quantity misrepresented,
or the marginal penalized output (MPO). Finally, line
delta in Fig. 2 graphs the audit probability, 8.
Fig. 3 graphs the determination of optimal misrepresentation at the industry level. The lines DELTA and
MPO are the horizontal summation of individual farmers' delta and MPO curves, respectively. Both curves
have an intercept of 8o when they are graphed relative
When the combination of policy variable and enforcement parameters are such that farmers misrepresent their production, traditional analysis fails to consider the area BEGH. This area represents farmers'
expected benefits from cheating, E[Bc] = [v- 8(v +
p)] Qm. The benefits from cheating constitute a decoupled income transfer from taxpayers to producers,
since the transfer does not affect farmers' production
decisions. 7
Even though not present in the stylized Fig. 3, monitoring and enforcement costs should be included in
both the taxpayers' costs and the welfare losses from
market intervention. The resource costs of enforcing
the program by investigating the farmers and convicting the detected cheaters, denoted as <1>(8o), are assumed to be a non-decreasing function of the base audit probability (i.e., <1>'(8o) 2:0, <1>"(8o) 2: 0).
There are also fixed costs associated with the operation of the enforcement agency. These costs are not
incorporated into the model. The reason for the exclusion of these fixed costs lies in the presumption
that the existence of the agency responsible for policy
enforcement depends on government intervention in
agriculture rather than the presence of any commodity
program in particular.
7 Due to the decoupled nature of the surplus transfer through
output misrepresentation, the analysis of cheating on output subsidies presented in this section of the paper is not affected by
any assumption about trade consideration. Consider the case of
an output subsidy adopted by a country that exports the regulated
commodity. When the country faces a perfectly elastic demand
curve (i.e., case of a small open economy), pc in Eq. (I) will account for the (unaffected) world price. In the large country case,
the market price (pc in Eq. (!)) would reflect the (distorted) world
price after intervention. In either case, the subsidy v equals the
difference between the price received by producers and the relevant world price and the best response function of the farmers is
the one given by Eq. (4).
K. Giannakas, M. Fulton/Agricultural Economics 22 (2000) 75-90
80
l>,mpo
Fig. 2. Optimal fanner misrepresentation on output subsidies (cheating equilibrium).
p
E
~~~~-~--,-~~~
(f~i·
;
L
~'
I
<
/
.v
DE!<TA
'.' "f~~~i ....
.......
L'-·-·---·-}--·-·-·-·~-.-~-- v/(v+p)
HI ./ )·o .-:-'~----·-· li (v+p)
~~:ih>'
p• .........
1;'fPO
--·--------~----
- Qm
- - - >Qm
Fig. 3. The welfare effects of output subsidies under costly enforcement and misrepresentation.
4. Optimal enforcement by the enforcement
agency
Eq. (4) indicates that farmer misrepresentation under a subsidy scheme depends on the level of the pay-
ment and the enforcement parameters. Since, however,
the enforcement parameters and the policy variable are
endogenous to agricultural policy makers, the question
that arises is whether complete deterrence of cheating
is the optimal response of regulatory and enforcement
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
agencies to the (optimizing) behavior of the farmers
(which has been shown to include cheating when allowed by the circumstances).
This section of the paper examines the problem
of policy enforcers. The problem of the enforcement
agency is to determine the degree to which the subsidy scheme designed by the regulator is enforced. In
making this decision, the enforcement agency knows
exactly how its decisions will affect the (optimizing)
behavior and welfare of farmers.
The level of enforcement is determined by the enforcement parameters, the audit probability and the
penalties. Penalties for producers detected cheating on
farm programs are generally set elsewhere in the legal
system and are, therefore, exogenous to agricultural
policy makers. With p exogenous to agricultural policy makers, the problem of the enforcement agency
is the determination of the 8o that maximizes its objective function. The general form of the enforcement
agency's problem can be written as:
max8 0 W
=e
I
= e PS +
TS
1
Q'
S(Q 1) Q 1 -
S(Q) dQ + [(1- 8) (S(Q 1)
-D(Q 1) ) - 8p]Qm ) - (1 +d) {[S(Q 1) - D(Q 1)]Q 1
+[(1- 8)(S(Q 1) - D(Q 1) ) - 8p]Qm + <l>(8o)}
s. t.
(5)
Qm __ v-8o(v + p)
28~(v+p)
where D(Q) and S(Q) are the inverse demand and
supply functions, respectively, and e is the weight
placed by the enforcement agency on producer welfare. All other variables are as previously defined. The
consumer surplus is not included in the objective function of policy enforcers since, for any output subsidy
v, cheating involves direct transfers from taxpayers to
producers. Thus, consumer welfare is not affected by
the amount of enforcement and farmer misrepresentation.
Assuming the enforcement costs <l>(8o) equal
1/21/185 (where 1/1 is a strictly positive scalar that
depends on things like the agrarian structure and
the number of representative farmers), the first order
condition for the problem specified in Eq. (5) is
aw
a80 = o =?
(1
81
+ d)1/18o = [(1 +d) - e]
. [S(Qt)- D(Qt) . [S(Qt)- D(Qt) + p J ]
28'
28'
8o
1
(6)
1
Eq. (6) indicates that the optimal audit probability
is determined by equating the marginal resource costs
of enforcement (MCe = (1 + d)1/f8o), with the marginal
benefits from investigation (MBe = [(1 +d) -8]((v8o (v + p))) /28~). The marginal benefits from enforcement include benefits from penalties on the current
level of misrepresentation and the benefits from increased enforcement and reduced cheating.
The effect of policy enforcement on farmers'
well-being may or may not be taken into account by
policy enforcers. For various reasons, the enforcement agency might place a relatively high weight
(1:1 =eH, where eH:::: 1 +d), a low weight (1:1 =eL,
where eL E(0, 1 +d)), or no weight (1:1 =e0 =O) on
producer surplus. Substituting these values into Eq.
(6) and solving for 8o generates the best response
function of the enforcement agency to the output subsidy chosen by the regulator and farmers' optimizing
behavior for the three values of e.
More specifically, when the enforcement agency
does not consider the effect of its choices on producers' welfare (i.e., 8 = 8° = 0), but its objective, instead,
is to minimize taxpayer costs from cheating, 8 the base
audit probability will equal
v
(7)
where the superscript denotes the weight placed by
policy enforcers on producer surplus. Similarly, when
the enforcement agency places a positive but relatively
low weight on producer surplus (i.e., e is lower than
the m~rginal cost of public funds, 1 +d), the optimal
8o, og , will equal
8 sub stltutmg
· ·
e0 mto
·
Eq. (5) shows that enforcement agency's
payoff function is measured by the addition to the regulator's
revenue net of enforcement costs. Alternatively, the enforcement
agency can be viewed as seeking the 8o that minimizes total budgetary costs from cheating, i.e., the resource costs of enforcement
plus the payments on quantity misrepresented minus the penalties
collected from those detected cheating.
82
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
Panel (a): Optimal Enforcement
MB.,MC.
Panel (b): Optimal Misrepresentation
I>,MPO
/
/
v
v+p
.·
.·
sg0 F-~---+-.r.-----~------------;r------~~r
/
l)gL ............... ~--···!················]······························· ··················
//
/
:
:
:
:
:
~
~
/
//
:
ligH 0~---a~o--8-L
:
_ __.e~H----•
Qm
Qm
Qm
Qm
Fig. 4. Optimal enforcement and strategic interdependence between the enforcement agency and the farmers.
0eL _
0
-
[(1 +d) - 8][S(Qt) - D(Qt)]
[(l+d)-8][S(Qt)-D(Qt)+pJ+(l+d)2811fr
[(1 +d) - B]v
[(1 +d)- 8](v + p)
+ (1 +d) 2811fr
(8)
When the weight placed on producers exceeds the
marginal cost of misrepresentation to taxpayers (i.e.,
8 2:: 1+d), the best response of policy enforcers is
complete allowance of cheating, i.e.,
(9)
A zero base audit probability does not mean that
cheating goes undetected. Since 81 is assumed to be
strictly positive, a zero oo means that policy enforcers
will not actively spend resources to deter misrepresentation over and above that which would occur otherwise.The reaction functions of policy enforcers under
the different B's indicate that Bo decreases with the increased Weight placed On producers (i.e., ogO > ogL >
ogH).
Maximum enforcement occurs when policy enforcers place zero weight on producer welfare. Enforcement, however, is always incomplete due to the
83
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
positive resource costs of investigation (i.e., 1f; > 0);
8o will be smaller than the base audit probability that
completely deters cheating (i.e., 8o < 80c = v / (v +
p)).
The optimal8o under the alternative political preferences of the enforcement agency is determined graphically by the intersection of the MCe curve with the
relevant MBe curve in Fig. 4, Panel (a). When e equals
zero, the relevant marginal benefit function is shown
as the downward sloping solid MBe curve. The MBe
curve is downward sloping due to the decrease in misrepresentation caused by increases in 8o. The intersection of the MBe curve with the horizontal axis determines the base audit probability that completely deters cheating, 80c. Obviously, 80cwould be the optimal
choice of policy enforcers if enforcement was costless
(i.e., 1f; = 0). In this case, the MCe curve would coincide with the horizontal axis. However, investigating
farmers and convicting the detected cheaters is costly.
The greater are the enforcement costs (i.e., the larger
is 1f; ), the greater is the slope of the MCe curve, and
the lower is the base audit probability. 9
An increase in the weight policy enforcers place on
producer surplus reduces both the intercept and the
absolute value of the slope of the marginal benefit
function. More specifically, increases in e cause a leftward rotation of the MBe curve through 80c. Ceteris
paribus, this results in a reduced base audit probability. Under eL, the relevant MBe curve (shown as the
downward sloping dashed MBe curve in Fig. 4, Panel
(a)) will always fall between the MBe curve under e0
and the horizontal axis; 8gL is always positive.
When e = 1 + d, the weight placed by policy enforcers on producers equals the marginal cost of
public funds, i.e. the implicit weight placed by the enforcement agency on taxpayer surplus. Since taxpayer
gains from increased enforcement constitute producer
losses (in a one-to-one correspondence) and since
equal weight is attached to the welfare of producers
and taxpayers, the marginal benefits from enforcement are zero. Hence, when e = 1 + d, the MBe curve
9
Although Panel (a) of Fig. 4 illustrates the case of increasing
marginal enforcement costs, the marginal costs from enforcement
can in fact be constant, i.e. the case of constant returns to government spending on program enforcement. In such a case, the
relevant Mce curve would be a horizontal line that would meet the
vertical axis in Panel (a) of Fig. 4 at 1/f', the level of the constant
marginal costs.
coincides with the horizontal axis, and both the slope
and the intercept equal zero. The only point where the
MCe curve meets the horizontal axis is at the origin.
Thus, the optimal 8o equals zero.
Finally, values of e greater than 1 + d result in a further leftward rotation of the MBe curve. The relevant
MBe curve is shown as the upward sloping dashed line
in Fig. 4, Panel (a). Since the weight placed on producers exceeds the marginal cost of public funds, the
benefits from investigating farmers are never positive.
Thus, when policy enforcement is costly (i.e., whenever 1f; > 0), the best response of policy enforcers that
place relatively high weight on producers is to choose
a zero base audit probability.
Fig. 4 also graphs the strategic interdependence between the enforcement agency and farmers; it shows
the effect enforcement decisions have on output misrepresentation. Panel (b) of Fig. 4 depicts the cheating
equilibrium for the N representative farmers. Changes
in 8o result in parallel shifts of the MPO curve faced
by the farmers. More specifically, reductions in 8o
caused by increases in e translate into downward parallel shifts of the MPO curve and increased output misrepresentation for a given subsidy and penalty. Mathematically, Qm under the different political preferences
of policy enforcers is derived by substituting the appropriate 8o into the farmers' reaction function in Eq.
(4 ). Hence, when e =e0 output misrepresentation will
equal
Qeo _
m -
1j;v
(V
+ p) ( V + p + 281 1/1)
(10)
Similarly, the equilibrium Qm under eL and eH, Q~
and Q~H, respectively, will equal
gL
Qm =
(1
+ d)1j;v
----------~----~~----------
(v
+ p )[(1 +d) (v+p+28~ 1/J)-e (v + p )]
(11)
and
gH
V
Qm = 28~ (v + p)
(12)
Fig. 4 is well suited for comparative static's analyses. For instance, an increase in the penalty results in
a parallel downward shift of the vl(v + p) line in Panel
(b) and a reduction in Qm (direct effect). An increased
p also results in a clockwise rotation of the relevant
84
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
p
pt - - - - - -
-·--.~~---·-··
.·
v
MP09 H
•••
/
/
/
A
/
/
/
-·-·-·-·-·-·- v/(v+p)
p•.
E
o·
Q
Fig. 5. The welfare effects of output subsidies under various levels of enforcement and cheating.
downward sloping MBe curve through the intercept in
Panel (a). The optirnal8o is reduced and Qm increases
(indirect effect). A change in the subsidy results in
parallel shifts of the v/(v+ p) line in Panel (b) (direct
effect on Qm), and also changes both the intercept and
slope of the MBe function in Panel (a) (indirect effect).
Overall, when program enforcement is costly, complete deterrence of cheating on output subsidies is
never optimal from an economic perspective. The optimal enforcement and, therefore, the optimal output
misrepresentation depend on the weight placed by
policy enforcers on producer surplus. Enforcement is
maximized and cheating is minimized when the enforcement agency minimizes total taxpayer costs from
cheating. When farmer welfare is weighted highly,
complete allowance of cheating is the optimal choice
of the enforcement agency and maximum misrepresentation the best response of the farmers.
5. Regulator and optimal intervention
Consider next the case of a regulatory agency in a
decentralized policy making environment that desires
to transfer a given surplus to producers of the regulated
commodity. Implicit in this formulation is the assumption that the regulator is not concerned with the distribution of resources within the farm sector; the purpose of government intervention is to transfer income
to the farmers. This assumption is consistent with the
assumption of homogeneity of producers adopted in
this paper.
The regulator's problem can be seen as the determination of the subsidy level that achieves the desired
income redistribution. Since the reaction functions of
all parties involved in agricultural policy design and
implementation are assumed to be common knowledge, the regulator knows exactly whether and how
her decision will affect the level of enforcement and
output misrepresentation.
Assume that the political preferences of the regulator result in a desire for an income transfer to producers given by the areas A+ B in Fig. 5. When policy enforcement is perfect and costless, the quantity
reported as eligible for government payments equals
the actual production level. In such a case, the optimal
choice of the regulator in terms of v that transfers areas
A+ B(= [(pc + v)Qt _ C(Qt)] _ [peQe _ C(Qe)])
K. Giannakas, M. Fulton/Agricultural Economics 22 (2000) 75-90
to producers will equal the difference between l and
pc shown in Fig. 5. The optimal subsidy under perfect
and costless enforcement is denoted as vpce.
When, however, enforcement of output subsidies is
costly, it will be incomplete and some output misrepresentation will always occur. The extent of misrepresentation depends on the weight policy enforcers place
on producer surplus. Because of this misrepresentation, there is always more than the desired surplus
transferred to producers under a subsidy payment set
at vpce. The excess transfer increases with the increase
in misrepresentation.
Fig. 5 shows output misrepresentation and the welfare effects of a given subsidy under the alternative
political preferences of the enforcement agency. The
greater is (), the lower is 8o, and the greater is Qm. The
lower is enforcement and the greater is misrepresentation, the greater is the total transfer to producers (i.e.,
payments for output produced plus benefits from misrepresentation) for any given subsidy level. Thus, for
the equivalent of area A + B to be transferred to producers, the regulator has to reduce the unit payment to
the level at which the total transfer to producers will
equal A+ B. Therefore, the optimal subsidy that transfers a given surplus to producers falls with the increase
· rmsrepresentatwn,
·
·
·
m
1.e.,
veH < veL < v0° < vpee .
A consequence of this is that consumer surplus falls
with an increase in cheating when the objective of the
regulatory agency is to transfer a given surplus to producers. The taxpayer costs associated with the specific
income redistribution are reduced by the amount foregone by consumers plus the change in the deadweight
welfare loss triangle, adjusted to account for the distortionary costs from taxation. Furthermore, the reduced
enforcement associated with increased weight on producers (and increased cheating) results in reduced enforcement costs incurred by taxpayers. 10
10 The effects of the reduction in the subsidy when cheating
occurs in an open economy framework are quite straight forward.
When farmer misrepresentation occurs in a small open economy,
the welfare of consumers remains unaffected. The reason is that
the domestic policy has no effect on the world price. However, the
change in the subsidy will affect the terms of trade for the country.
More specifically, the reduction in the quantity produced will
reduce (increase) exports (imports) for the small open economy
that exports (imports) the supported commodity. For the large
country case, the reduction in domestic production (due to the fall
in output subsidy) will increase the world price. Increased world
85
6. Transfer efficiency and optimal income
redistribution
Assuming that the sole purpose of market intervention is income transfer, the trade off between producer surplus and consumer plus taxpayer surplus under an output subsidy scheme is reflected by the surplus transformation curve (STC) (Josling, 1974; Gardner, 1983; Bullock, 1992). The slope of the STC, denoted as s = aPSja(CS + TS), is the marginal rate
of surplus transformation. It reflects the efficiency of
the policy mechanism in redistributing income to producers at the margin, or how much of an extra dollar
raised by consumers and taxpayers is received by producers. One minus the absolute value of s shows the
deadweight loss per dollar transfer. The efficiency in
redistribution links the resource costs of market intervention to the surplus transferred to producers. The
closer is s to -1, the smaller are the welfare losses,
and the more efficient is the income redistributional
mechanism.
The analysis in the previous sections shows that the
levels of intervention and enforcement vary with the
political preferences of the enforcement agency. The
same is also true for the welfare loss associated with
a given transfer to producers. Variation in the social
cost of a transfer implies variation in the transfer efficiency of output subsidies. In general, less auditing
means lower resource costs of monitoring and enforcement associated with the specific transfer to producers. At the same time, the lower is the subsidy level
that achieves the desired transfer, the lower are the distortionary costs of market intervention (i.e., the Harberger triangle and the deadweight losses from taxation). And the lower are the welfare losses from a
given transfer to producers, the greater is the transfer
efficiency of output subsidies.
Recall that when policy enforcement is costly, both
enforcement and the subsidy decrease with an increase
in the weight placed by policy enforcers on producer
price means reduced surplus for domestic consumers. When the
large country is an exporter of the commodity, reduced production
means also reduced exports and therefore, reduced transfer from
domestic taxpayers to foreign consumers while the same surplus
is transferred to domestic producers. Similarly, when the large
country imports the subsidized commodity, the reduced subsidy
results in increased imports and reduced transfer from foreign
producers to domestic consumers.
86
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
PS
E
0
CS+TS
Fig. 6. Surplus transformation curves for output subsidies under costly enforcement.
.
o&O
o&L
gH
gO
gL
gH
surpl us (1.e., o 0 > o 0 > 80 and v > v > v ).
Since both enforcement costs and distortionary costs
of intervention decrease with an increase in e, the
greater is e, the greater is the marginal efficiency of
output subsidies in transferring income to producers,
i.e.lseH I > lseL I > Isea I·
Put in a different way, the efficiency of a transfer
to producers increases with cheating since output misrepresentation allows the regulator to substitute distortionary transfers through the market with more efficient decoupled transfers through cheating. Since enforcement falls as the weight placed by the enforcement agency on producer welfare increases, the greater
is e, the greater is the decoupled transfer through
cheating, and the more closely output subsidies approximate a lump-sum transfer program.The relevant
STes are depicted as the (concave) STeeH, STeeL
and STe 80 in Fig. 6. The STes originate from point
E which is the locus of the interest group surpluses
at the competitive equilibrium. 11 STe 80 lies under11 As long as there is government intervention in any other commodity market, the taxpayer surplus corresponding to point E incorporates the costs associated with the operation of the enforce-
neath STeeL which, in tum, lies underneath STe 8 H
everywhere to the left of E. 12 The vertical distance
between the STes reflects the difference in the welfare
losses associated with the specific transfer to producers under the different es. Since og 0 , .sgL and the distortionary costs of market intervention increase with
an increase in the level of intervention, the vertical
distance between the STes increases with a leftward
move from E.
It is worth noting that the efficiency of the policy mechanism in transferring income to producers
increases when the political preferences of the enforcement agency and the regulator coincide (i.e.,
when both agents place a relatively high weight on
producer welfare). Paradoxically, the transfer efficiency of output subsidies falls when the objective
of the enforcement agency is to minimize total taxpayer costs from cheating. The reason is the relament agency. When non-intervention in the specific market also
means non-intervention in the whole agricultural sector, taxpayer
surplus at point E would shift to the right by an amount equal to
the fixed costs associated with the existence of the enforcement
agency.
12 Obviously for zero subsidy (no intervention), neither enforcement nor misrepresentation will emerge.
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
tively high enforcement and intervention that occur
when policy enforcers place zero weight on producer
welfare.
The relative position of the STC for output subsidies in a world where program enforcement is perfect
and costless (i.e., STCpce) is determined solely by
the distortionary costs of market intervention (i.e.,
the relevant Harberger triangle and the deadweight
losses from taxation to finance the transfer). The
marginal efficiency of redistribution of output subsidies, lspcel, will always be less than lseH and STcrce
I,
will always lie underneath STCeH everywhere to the
left of E. The reasoning goes as follows. Because of
misrepresentation that occurs under eH, the subsidy
that achieves the desired transfer to producers will be
smaller than the subsidy under perfect and costless enforcement, i.e. vrce > veH. Reduced subsidy implies
reduced distortionary costs of intervention. Reduced
welfare losses associated with a given transfer to producers mean increased transfer efficiency of output
subsidies.
The position of STCpce relative to STeeL and
STCe 0 is case specific and depends on market conditions and the resource costs of enforcement. Even
though the distortionary costs of market intervention
are lower under tJL and tJ 0 (since VeL and yeO are
smaller than vrce), the enforcement costs are greater
than those under perfect and costless enforcement of
the program. The relative position of the STCs depends on the relative size of the total costs. For given
market conditions, the greater is <P(8o), the greater
is the likelihood that STCpce will lie above STeeL
and STCe 0 • Alternatively, for relatively low enforcement costs, STCpce will lie underneath STCeL and
STCe 0 •
The STC framework developed above can be used
to determine the socially optimal total transfer to producers. Suppose the problem faced by the regulatory
agency is the determination of income redistribution
that maximizes some social welfare function (SWF)
that weights producer, consumer, and taxpayer welfare, rather than the determination of the subsidy level
that transfers a given surplus to producers. Assume
that the political preferences of the regulator result in
the social indifference curves (SIC) shown in Fig. 6,
with the SWF value increasing with a northeast shift
of the SIC.
87
The socially optimal total transfer to producers under the different scenarios considered in this study is
determined by the tangency of the SIC to the relevant
STC (Gardner, 1987). Fig. 6 shows that the level of
total transfer to producers is directly related to the efficiency of output subsidies in transferring income to
producers. More specifically, the greater is the transfer efficiency of the policy instrument, the larger is the
socially optimal total transfer. Furthermore, since the
SWF value increases with movements to the northeast,
increases in transfer efficiency also imply increases in
social welfare. Both the socially optimal total transfer
to producers and the social welfare from intervention
are maximized when the political preferences of the
enforcement agency and the regulator coincide.
7. Extension of the model- endogenous penalties
Crucial for the previous analysis and results is the
assumption that penalties are exogenous to agricultural policy makers. Consider now the case of an enforcement agency that controls both enforcement parameters - 8o and p. Endogeneity of penalties calls
for an additional first order condition to the enforcing
agency's problem specified in Eq. (5), i.e.,
aw = 0 ==? [(1 +d) ap
-
tl]
[ 85
[S(Qt)- D(Qt)p
]
. 48i - [S(Qt)- D(Qt) + p] 2 48l
= 0 ==? p =
(1 - 8o)
---v
8o
(1- 8o) [S(Qt)- D(Qt)]
8o
(13)
The optimal pin Eq. (13) is the penalty structure required to completely deter farmer misrepresentation,
pnc. Interestingly enough, the best response function
of the enforcement agency does not depend on the
weight it places on producer surplus. Solving Eqs.
(7)-(9) simultaneously with Eq. (13) indicates that
when penalties are endogenous to agricultural policy
makers, cheating will be completely deterred by a zero
base audit probability and a huge penalty on detected
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
88
misrepresentation. 13 This is true no matter the weight
placed on producers, i.e.,
o~(p)
=0
and
p
= oove
(14)
where 8~(p) denotes the optimal 8o under all es when
penalties are endogenous.
Graphically, the huge per unit penalty makes the
slope of all MBe curves in Fig. 4, Panel (a) infinite.
The MBe curves coincide with the vertical axis and
meet the MCe curve at the origin. The resulting zero
8o means that the MPO curve comes out from the
origin, while the huge penalty shifts the v/(v + p) line
downwards so that it coincides with the horizontal axis
in Fig. 4, Panel (b). The optimal response of the farmer
is then a zero level of misrepresentation.
Assuming there are no costs associated with the establishment of huge penalties on detected output misrepresentation, and since no (costly) auditing prevails
at equilibrium, the perfect enforcement of the program
(i.e., Eq. (13)) is also costless. Since output misrepresentation is perfectly and costlessly deterred when
penalties are endogenous to the enforcement agency,
the output subsidy that transfers a given surplus to
producers, the transfer efficiency of the policy instrument, and the socially optimal total transfer to producers are those derived by the traditional analysis of output subsidies. Thus, one interpretation of 'perfect and
costless enforcement' is the costless establishment of
infinite per unit penalties for farmers who are detected
misrepresenting the level of their production.
Infinite per unit penalties for farmers cheating on
farm subsidies is not what is observed in most of today's world however. In most countries legal penalties cannot be set far in excess of the material damage
caused by the illegal activity; in lay terms, the punishment has to fit the crime. In the EU for instance, detected farmer misrepresentation on compensatory payments results in a scaling-back of payments by the
13 This result is altered when agricultural policy making is centralized and penalties are endogenous to agticultural policy makers.
Giannakas (1998); Giannakas and Fulton (1999) show that when
a single agency determines both the level of intervention (i.e., the
subsidy) and the level of enforcement, policy makers will find it
economically optimal to completely allow cheating. The reason is
that the increased producer benefits from misrepresentation enable
the singe agency's policy makers to reduce the subsidy that transfers a given surplus to producers, thereby increasing the transfer
efficiency of the policy.
same percentage (i.e., the percentage of cheating) plus
an additional uncompensated set-aside requirement for
the following year (Swinbank and Tanner, 1996, pp.
97). 14 Even in the US where detection of cheating on
farm income transfer programs causes punitive fines
and exclusion from future program benefits (usually
for 3 years), the penalty is not of the magnitude that
results in costless policy enforcement, i.e., the per unit
penalty is not infinite.
One reason why the severe punishment of law violators is not the norm is that institutionalized infinite
per unit penalties are neither credible nor just. For
instance, in his work on income tax evasion Cowell
(1990, pp. 150) argues that enormous fines for small
amounts of cheating would violate 'the public's sense
of what is fair and reasonable.' Similar views can be
found in Becker (1968); Stigler (1970); Polinsky and
Shaven (1979); Shaven (1987).
8. Summary and discussion
Agricultural policy analysis, in general and the analysis of output subsidies in particular, traditionally take
place under the assumption that: (i) either farmers
do not cheat; or (ii) enforcement of agricultural policies is perfect and costless. However, enforcement requires resources and is, therefore, costly. The resource
costs of enforcing an output subsidy scheme result in
policy enforcement that is incomplete. Imperfect enforcement generates economic incentives for farmers
to misrepresent their production and to collect subsidies on production that never took place. The lower is
the level of enforcement, the higher is the equilibrium
amount of farmer misrepresentation.
The analysis in this paper shows that the level of
enforcement depends on the political preferences of
policy enforcers. Since cheating on subsidies results
in a direct income transfer from taxpayers to produc14 In 1993 Germany successfully resisted the enforcement of the
rules when in parts of Eastern Germany the areas reported as
eligible for payments exceeded the base areas by up to 15%. Not
only were producers not penalized, but Germany also managed
to exclude these areas from the penalty scheme for the following
3 years. The decision of the European Council to not punish the
German producers created a precedent in the EU and resulted in
detected cheating being unpunished in other parts of the Union as
well (e.g., Spain and Scotland) (Swinbank and Tanner, 1996).
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
ers, enforcement decreases as policy enforcers place
increasing weight on producer welfare. The political preferences of policy enforcers are also crucial
in determining the subsidy that transfers a given surplus to producers, the transfer efficiency of the policy mechanism, and the socially optimal total transfer to producers. The causation goes as follows. The
greater is the weight placed by policy enforcers on
producer surplus, the lower is the program enforcement, and the lower is the subsidy level that achieves a
desired transfer to producers. Lower enforcement and
intervention means lower welfare losses associated
with a given transfer to producers, and greater transfer efficiency of output subsidies. The greater is the
marginal efficiency of output subsidies in redistributing income to producers, the greater is the socially
optimal income redistribution. Overall, the transfer efficiency of output subsidies, the socially optimal total transfer to producers and the social welfare from
intervention are maximized when the political preferences of the enforcement agency and the regulator
~oincide.
When both the enforcement agency and the regulator attach a relatively high weight to the welfare
of producers, the economically efficient outcome includes a reduced subsidy and a minimum amount of
monitoring and enforcement. The conclusion that a
limiting case of the analysis approximates a lump-sum
transfer highlights at least two issues in the literature
on lump-sum payments. First, even though lump-sum
transfers are more efficient means of income redistribution than coupled farm subsidies, they are viewed
as 'hypothetical ideals' that 'are operationally irrelevant' (Williamson, 1996, pp. 210). Second, lump-sum
transfers are not only impossible, but they are perceived to be unfair to the extent that they involve arbitrary exclusions from the program (Gardner, 1987,
pp. 190).
Lump-sum transfers that are linked to cheating obviously involve highly arbitrary exclusions. Farmers
that comply with program provisions and truthfully report their production receive less benefits than do the
farmers that cheat. While the focus on the representative farmer in this paper precludes the formal consideration of this outcome, the principle is nevertheless
clearly illustrated.
The impossibility of lump-sum transfers is also
highlighted by the analysis in this paper. While con-
89
ceptually interesting, the idea that cheating would be
made the basis of policy seems far-fetched. One reason is that morality and culture, although significant
determinants of an individual's propensity to cheat
(Grasmick and Green, 1980), have not been incorporated into the analysis. Not only are some farmers expected not to cheat, but wide spread cheating is likely
to create a culture of dishonesty in the society and a
public disrespect for both the government and community rules (Lea et al., 1987; Cowell, 1990). As a
consequence, governments would be unlikely to base
policy on something that would ultimately undermine
policy. The key point here is that any lump-sum transfer program that could be devised is predicated on a set
of behavioral assumptions that make it operationally
impractical.
The above discussion sheds light on the government's use of mixed policy instruments such as
subsidies and supply restrictions (i.e., the so-called
stop-and-go policies). Linking subsidy payments to
an arbitrary chosen level of output (or input) does two
things. First, it provides a way of making operational
a program that now has many of the characteristics
of a lump-sum transfer. Second, it may reduce the
possibility of cheating, thus limiting the exclusion
of honest farmers and lessening the general propensity for cheating. The possibility of cheating may
be reduced because the enforcement issues in this
case are confined to compliance with supply restrictions which may be easier to monitor than output
(on the economics of a stop-and-go policy mix under costly enforcement see Giannakas and Fulton
(1999)).
In addition to providing an understanding of the
causes and consequences of cheating on output subsidies, the results of this study might assist in explaining potential differences in compliance with
policy rules observed in different areas/countries.
For instance, the varying significance of agriculture in different countries, and the different weights
being placed by Ministries of Agriculture or agricultural policy enforcement agencies on producer
welfare, might account for the variable levels of
enforcement and cheating seen around the world.
More research, however, is required to analyze and
better understand these issues and to determine the
empirical importance of cheating on output subsidies.
90
K. Giannakas, M. Fulton/ Agricultural Economics 22 (2000) 75-90
References
Alston, J.M., Hurd, B.H., 1990. Some neglected social costs of
government spending in farm programs. Am. J. Agric. Econ.
72, 149-156.
Alston, J.M., Smith, V.H., 1983. Some economic implications
of minimum pricing: the case of wine grapes in Australia:
comment. Rev. Marketing Agric. Econ. 51, 179-185.
Ballard, C.L., Fullerton, D., 1992. Distortionary taxes and the
provision of public goods. J. Econ. Persp. 6, 117-131.
Becker, G.S., 1968. Crime and punishment: an economic approach.
J. Polit. Econ. 76, 169-217.
Bullock, D.S., 1992. Redistributing income to European consumers
and taxpayers through the common agricultural policy. Am. J.
Agric. Econ. 74, 59-67.
Cowell, F.A., 1990. Cheating the Government: The Economics of
Evasion. The MIT Press, Cambridge, MA, 267 pp.
From, J., Stava, P., 1993. Implementation of community law: the
last stronghold of national control? In: Andersen, S.S., Eliassen,
K.A., (Eds.), Making Policy in Europe: The Europification
of National Policy-making. SAGE Publications, London, pp.
55-67.
Gardner, B.L., 1983. Efficient redistribution through commodity
markets. Am. J. Agric. Econ. 65, 225-234.
Gardner, B.L., 1987. The Economics of Agricultural Policies.
Macmillan, New York, 387 pp.
Gardner, B., 1996. European Agriculture: Policies, Production and
Trade. Routledge, New York, 233 pp.
Giannakas, K., 1998. Agricultural policy analysis under costly
enforcement: an economic analysis of cheating, Ph.D.
Dissertation. University of Saskatchewan.
Giannakas, K., Fulton, M.E., 1999. Efficient redistribution using
quotas and subsidies in the presence of misrepresentation and
cheating, Am. J. Agric. Econ. (forthcoming).
Gibbons, R., 1992. Game Theory for Applied Economists.
Princeton University Press, Princeton, NJ, 267 pp.
Grasmick, H.G., Green, D., 1980. Legal punishment, social
disapproval and internalization as inhibitions of illegal behavior.
J. Crim. Law Criminal. 71, 325-335.
Harvey, D.R., 1997. Extensions and political analysis of the CAP.
In: Ritson, C., Harvey, D.R., (Eds.), The Common Agricultural
Policy. CAB International, New York, pp. 163-190.
Josling, T.E., 1974. Agricultural policies in developed countries:
a review. J. Agric. Econ. 25, 229-264.
Kreps, D.M., 1990. A Course in Microeconomic Theory. Princeton
University Press, Princeton, NJ, 839 pp.
Lea, S.E.G., Tarpy, R.M., Webley, P., 1987. The Individual In The
Economy. Cambridge University Press, Cambridge, 627 pp.
Moyer, H.W., Josling, T.E., 1990. Agricultural Policy Reform:
Politics and Process in the EC and the USA. Iowa State
University Press, Ames, lA, 235 pp.
Nerlove, M., 1958. The Dynamics of Supply: Estimation of
Farmers' Response to Price. Johns Hopkins Press, Baltimore,
268 pp.
Ockenden, J., Franklin, M., 1995. European agriculture: making
the CAP fit the future. Royal Institute of International Affairs,
London, UK, 103 pp.
Polinsky, M., Shavell, S., 1979. The optimal trade-off between
the probability and magnitude of fines. Am. Econ. Rev. 69,
880-891.
Runge, C. F., von Witzke, H., 1987. Institutional change in the
common agricultural policy of the European Community. Am.
J. Agric. Econ. 69, 213-222.
Shavell, S., 1987. The optimal use of nonmonetary sanctions as a
deterrent. Am. Econ. Rev. 77, 584-592.
Stigler, G.J., 1970. The optimum enforcement of laws. J. Polit.
Econ. 78, 526-536.
Swinbank, A., 1997. The CAP Decision-making Process. In:
Ritson, C., Harvey, D.R., (Eds.), The Common Agricultural
Policy. CAB International, New York, pp. 57-80.
Swinbank, A., Tanner, C., 1996. Farm Policy and Trade Conflict:
The Uruguay Round and CAP Reform. The University of
Michigan Press, 178 pp.
United States Department of Agriculture Office of Inspector
General, July 1999. USDA Fraud Hotline www.usda.gov/oigl
hotline. htm.
Wallace, T.D., 1962. Measures of social costs of agricultural
programs. J. Farm Econ. 44, 580-594.
Williamson, O.E., 1996. The politics and economics of
redistribution and inefficiency. In: The Mechanisms of
Governance. Oxford University Press. Oxford, UK, pp.
195-213.